Why Did ACX Reject Your Audiobook? 6 Common Reasons (and Fixes)

ACX rejects audiobooks for six common reasons: a noise floor above -60 dBFS, peaks over -3 dBFS, RMS outside the -23 to -18 dBFS window, missing head and tail room tone, missing or wrong opening and closing credits, and a sample or bit rate other than 44.1 kHz 192 kbps CBR MP3. Most are fixable in your editor.

An ACX rejection is almost always a technical fault rather than a verdict on your narration. ACX runs an automated audio check on every uploaded file that measures hard numbers — noise floor, peak level, RMS loudness, sample rate, and bit rate — before any human listens to the book, so a file that misses one threshold is returned even when the reading is excellent. This guide names the six rejection reasons authors hit most often and gives the concrete fix for each. For the underlying numbers in full, see our companion guide to ACX audio requirements and submission specs; this post is the failures-and-fixes counterpart to it.

Why does ACX reject audiobooks in the first place?

ACX rejects audiobooks to keep a consistent listening experience across Audible's catalogue, enforced with an automated technical check plus a human quality review. The automated check measures each file against fixed thresholds for loudness, peak level, noise, format, and structure; the human review then listens for performance issues such as mouth noise and mispronunciations. A rejection notice tells you which files failed and, for the automated check, which threshold they missed. The official thresholds live on the ACX audio submission requirements page, the primary source for every number in this article. It helps to split the six reasons below into level faults (noise floor, peaks, RMS), fixed in mastering, and structural faults (room tone, credits, format, consistency), fixed by editing or re-exporting.

Reason 1: Your noise floor is louder than -60 dBFS

A noise floor louder than -60 dBFS is one of the most frequent ACX rejection reasons, and it comes from the quiet hiss in your recording rather than your voice. The noise floor is the level of background sound during silence — room hum, computer fans, traffic, microphone self-noise — and ACX requires it to sit below -60 dBFS. An untreated room or a noisy preamp pushes the floor above that line and the file fails.

The fix is to lower the noise at the source first and clean up the rest in software. Record in the quietest, most soft-furnished space you have, turn off fans and appliances, and move the microphone closer so you can record at a lower gain. Then apply gentle noise reduction in your editor (Audacity, Reaper, or similar) using a sample of your room tone as the noise profile — but avoid over-processing, because aggressive reduction creates a watery, artefacted sound the human reviewer will flag separately.

Reason 2: Your peaks exceed -3 dBFS (clipping)

Peaks above -3 dBFS are an automated-check failure because ACX requires the loudest moment of every file to stay at or below -3 dBFS, leaving headroom so the audio never clips. Recording too hot — input gain so high that loud syllables, plosives, or laughs slam into 0 dBFS — produces clipped peaks that sound harsh and distorted. The check measures true peak, so even brief inter-sample overshoots can push a file over the line.

The fix is to limit or normalize your finished file to a -3 dBFS ceiling, then re-export. Record with comfortable headroom (peaks landing around -6 to -10 dBFS), then apply a limiter at -3 dBFS during mastering as a safety net. Do not simply turn an already-clipped track down, because clipping distorts the waveform permanently and lowering the volume keeps the distortion. If peaks were clipped on the way in, the cleanest fix is to re-record that passage with lower gain.

Reason 3: Your RMS is outside the -23 to -18 dBFS window

RMS loudness outside the -23 to -18 dBFS range is a rejection reason because ACX requires the average level of each file to land inside that narrow window. RMS measures perceived loudness over time rather than the single loudest peak, so a file that is too quiet (below -23 dBFS) forces listeners to crank the volume, while one too loud (above -18 dBFS) feels fatiguing and sits awkwardly against the rest of Audible's catalogue.

The fix is to measure RMS per file and adjust gain to centre it around -20 dBFS before final export. Use your editor's RMS or loudness meter on each chapter, then apply a flat gain change to bring the average into the window. Light compression beforehand evens out your loudest and quietest delivery, making the RMS window easier to hit without pushing peaks past -3 dBFS. Aiming for the middle of the range gives margin against small file-to-file variation.

Reason 4: You are missing room tone at the head and tail

Missing room tone at the start and end of a file is a structural rejection reason that surprises first-time narrators. ACX requires roughly 0.5 to 1 second of room tone — the natural ambient sound of your recording space with no speech — at the head and tail of every file, so transitions sound like a continuous space rather than abrupt digital silence. Authors who delete all the quiet parts, or paste in true digital silence, fail this because the dead-air gap is audibly jarring.

The fix is to record a clean stretch of room tone and place it at the beginning and end of each file. Sit silent for several seconds at the start of a session to capture that ambience, then copy a short piece of it to the head and tail of every chapter. Use recorded room tone, not your editor's "insert silence" function, because digital silence drops to absolute quiet — exactly what the requirement exists to prevent.

Reason 5: Your opening or closing credits are missing or wrong

Missing or incorrect opening and closing credits are a rejection reason because ACX requires every audiobook to begin and end with specific spoken credits. The opening must state the book title, the author, and the narrator; the closing must state that the book is finished and, where applicable, repeat key credits. Many first submissions are returned because the narrator jumped straight into Chapter 1 with no opening credit, or ended on the book's last line with no closing line.

The fix is to record dedicated opening and closing credit segments as the first and last sections of the audiobook. Script the opening as "[Title], written by [Author], narrated by [Narrator]," and the closing to confirm the book is complete. Check the exact wording ACX expects on the ACX submission requirements page before recording, because missing one required element fails the whole credit check.

Reason 6: Your sample rate or bit rate is wrong

A sample rate or bit rate that does not match ACX's format is an automated-check failure with nothing to do with how the book sounds. ACX requires each file to be a 192 kbps or higher constant bit-rate (CBR) MP3 at a 44.1 kHz sample rate. Files exported at 48 kHz, at a variable bit rate (VBR), or below 192 kbps are bounced even when the loudness and noise numbers are perfect, because the container itself is out of spec.

The fix is to set export settings correctly before rendering. In your editor's export dialog, choose MP3, set the sample rate to 44.1 kHz, select constant bit rate, and set 192 kbps. If you mastered at 48 kHz, resample the project to 44.1 kHz rather than just relabelling the file. Exporting every chapter from a single saved preset prevents one stray file slipping out at the wrong setting — which leads directly to the seventh issue worth naming.

What about consistency across files?

Inconsistency across files is a rejection reason that sits on top of the other six, because ACX expects every file in a single audiobook to match in volume, tone, pacing, and format. A book where Chapter 3 was recorded on a different day, room, or gain can have one file that drifts below the RMS window or above the noise floor even though the others pass. The human reviewer also flags audible shifts in microphone distance or vocal energy between chapters.

The fix is to master every file through the same chain and check them as a set, not one at a time. Apply the same noise reduction, compression, limiting, and RMS target to every chapter from a saved template, and export all files from one preset so the format never varies. Before submitting, listen to the join between a few chapter ends and the next chapter's start to confirm the room and voice match. This pass is where a second set of ears helps — part of why a human-review step matters so much in production.

ACX rejection reasons and fixes at a glance

The table below summarises each common ACX rejection reason, the exact threshold or requirement behind it, and the practical fix. Read it as a checklist to run against every file before you submit, working top to bottom so level faults are corrected before you re-export and re-check the format.

The key takeaway from the table is that almost every ACX rejection is a mastering or assembly problem rather than a narration problem, so fixing it usually means re-processing and re-exporting the flagged files rather than re-recording the whole book. Level faults (noise floor, peaks, RMS) are fixed in the master; structural faults (room tone, credits, format, consistency) are fixed in the file. If you correct the flagged files and re-run this checklist on the full set, the resubmission almost always passes.

How does TomeVox handle these ACX failure points?

TomeVox masters every audiobook to ACX's published specs, so the six failure points above are handled before delivery rather than discovered after a rejection. Each finished file is produced with a noise floor below -60 dBFS, peaks at or below -3 dBFS, RMS inside the -23 to -18 dBFS window, room tone at the head and tail, and 44.1 kHz 192 kbps CBR MP3 format, and every audiobook is automatically checked for technical quality before it ships. Because the whole book runs through one consistent chain, the files match each other in level and tone — the consistency requirement that trips up multi-session DIY recordings.

There is an honesty point about where a TomeVox file can go, because mastering to ACX specs is not the same as being eligible for the standard ACX portal. Standard ACX submission requires human narration unless otherwise authorized; Audible has announced acceptance of third-party AI narration but it is not yet open self-service to all indie authors. So an author has two paths: use ACX with human narration (this checklist is your pre-flight), or distribute a TomeVox AI-narrated file through the AI-friendly channels instead. A TomeVox file uploads directly to Google Play Books and Kobo, and reaches Apple Books, Spotify, and Chirp through an aggregator that accepts AI narration such as PublishDrive or Author's Republic — with AI/digital-voice disclosure required everywhere. Our guide to where to sell an AI audiobook maps every channel, and the AI audiobook commercial rights guide covers ownership.

The trade-off is narration source against distribution reach. To be on Audible through standard ACX today you need a human narrator, and the checklist above keeps that file from being rejected. For a finished, mastered, fully owned AI file fast, TomeVox delivers an M4B with chapter markers plus per-chapter MP3 files, usually within 48 hours, for a flat $49 to $99, distributed wide through the AI-friendly channels. The best AI audiobook tools comparison puts the production options side by side, and the cheapest way to make an audiobook guide weighs cost across every route.

Frequently asked questions

What is the most common reason ACX rejects an audiobook?

The most common technical reasons ACX rejects an audiobook are a noise floor louder than -60 dBFS and peak levels above -3 dBFS. A noisy room or a hissy microphone pushes the noise floor too high, and recording too hot causes peaks to clip. Both are caught by ACX's automated audio check before a human ever listens, so a file that fails either threshold is bounced back regardless of how good the performance is.

What are the ACX audio level requirements?

ACX requires each file to measure between -23 dBFS and -18 dBFS RMS, peak no higher than -3 dBFS, and a noise floor below -60 dBFS. Files must be 192 kbps or higher constant bit-rate MP3 at 44.1 kHz, include room tone, and carry opening and closing credits. A file outside any of these thresholds fails the automated check and is returned for correction.

What is room tone and why does ACX require it?

Room tone is a short stretch of the recording space's natural silence with no speech, used so a listener hears consistent ambience instead of abrupt digital silence between sections. ACX requires roughly 0.5 to 1 second of room tone at the start and end of every file. Pasting in true digital silence instead of recorded room tone is a common cause of rejection because the dead-air gap is audible and jarring.

Does ACX accept AI-narrated audiobooks?

Standard ACX submission requires human narration unless otherwise authorized. As of 2026, Audible has announced acceptance of third-party AI narration but it is not yet open self-service to all indie authors. If you produce an AI-narrated file with a tool like TomeVox, you distribute it through AI-friendly channels such as Google Play Books and Kobo directly, or Apple Books, Spotify, and Chirp through an aggregator, rather than through the standard ACX portal.

Do I have to resubmit my whole audiobook if one file fails?

No. ACX flags the specific files that fail its checks, so you only need to correct and re-upload the files identified in the rejection notice. However, because consistency across files is itself a review criterion, it is worth re-checking that the corrected files still match the volume, pacing, and tone of the rest of the book before resubmitting.